Distributed transactions and 2PC in DBMS Theory - Step-by-Step Execution

Concept Flow - Distributed transactions and 2PC

Client starts transaction

↓

Coordinator sends PREPARE to all participants

↓

Participants vote: YES or NO

↓

Coordinator receives all YES

↓

Coordinator sends COMMIT

↓

Participants commit

↓

Transaction ends

The coordinator asks all participants if they can commit. If all agree, it tells them to commit; if any say no, it tells them to abort.

Execution Sample

DBMS Theory

Coordinator: send PREPARE
Participants: vote YES or NO
Coordinator: if all YES send COMMIT else ABORT
Participants: commit or abort

This shows the basic message flow of the two-phase commit protocol in a distributed transaction.

Analysis Table

Step	Action	Coordinator State	Participants State	Message Sent	Result
1	Client starts transaction	Waiting	Idle	None	Transaction begins
2	Coordinator sends PREPARE	Waiting for votes	Ready to vote	PREPARE	Participants prepare to vote
3	Participants vote YES	Collecting votes	Vote YES	YES	Vote recorded
4	Participants vote YES	Collecting votes	Vote YES	YES	Vote recorded
5	Coordinator receives all YES	All votes YES	Waiting for commit/abort	None	Ready to commit
6	Coordinator sends COMMIT	Sent COMMIT	Received COMMIT	COMMIT	Participants prepare to commit
7	Participants commit	Waiting for ack	Committed	ACK	Commit done
8	Coordinator receives ACKs	Transaction complete	Idle	None	Transaction ends

💡 All participants voted YES, so coordinator sent COMMIT and transaction completed successfully.

State Tracker

Variable	Start	After Step 2	After Step 5	After Step 7	Final
Coordinator State	Waiting	Waiting for votes	All votes YES	Sent COMMIT	Transaction complete
Participants State	Idle	Ready to vote	Vote YES	Committed	Idle
Messages Sent	None	PREPARE	YES votes	COMMIT	ACK

Key Insights - 3 Insights

Why does the coordinator wait for all participants to vote before deciding?

Because the coordinator must ensure all participants agree to commit; if any say NO, the transaction must abort. This is shown in execution_table rows 3-5.

What happens if any participant votes NO?

The coordinator sends ABORT to all participants to cancel the transaction. This is not shown in the current table but would happen instead of sending COMMIT at step 6.

Why do participants send ACK after committing?

To inform the coordinator that they have completed the commit, so the coordinator knows the transaction is fully done. See execution_table rows 7-8.

Visual Quiz - 3 Questions

Look at the execution_table at step 5, what is the coordinator's state?

A All votes YES

B Waiting for votes

C Sent COMMIT

D Transaction complete

💡 Hint

Check the 'Coordinator State' column at step 5 in the execution_table.

At which step do participants commit the transaction?

A Step 4

B Step 6

C Step 7

D Step 8

💡 Hint

Look for 'Participants commit' in the 'Action' column of the execution_table.

If one participant votes NO at step 3, what message will the coordinator send next?

A PREPARE

B ABORT

C COMMIT

D ACK

💡 Hint

Recall the coordinator sends ABORT if any participant votes NO, as explained in key_moments.

Concept Snapshot

Distributed transactions involve multiple databases working together.
Two-Phase Commit (2PC) ensures all agree to commit or abort.
Phase 1: Coordinator asks participants to prepare and vote.
Phase 2: Coordinator sends COMMIT if all YES, else ABORT.
Participants commit or abort accordingly.
This prevents partial updates and keeps data consistent.

Full Transcript

Distributed transactions involve multiple databases or systems working together to complete a single transaction. The Two-Phase Commit protocol (2PC) is a method to ensure all parts agree to either commit or abort the transaction, keeping data consistent. First, the coordinator sends a PREPARE message to all participants asking if they can commit. Each participant votes YES or NO. If all vote YES, the coordinator sends a COMMIT message, and participants commit the transaction. If any vote NO, the coordinator sends ABORT, and participants cancel the transaction. Participants send acknowledgments after committing or aborting. This process prevents partial updates and ensures all systems stay in sync.